Finite Data Performance of MVDR Beamformer with and without Spatial Smoothing

advertisement
I
IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 40, NO. I I , NOVEMBER 1992
2126
Finite Data Performance Analysis of MVDR
Beamformer with and without Spatial
Smoothing
Kalavai J. Raghunath, Student Member, IEEE, and V. Umapathi Reddy, Senior Member, IEEE
Abstract-Recently, the performance of a minimum variance
distortionless response (MVDR) beamformer has been extensively studied for the case when a true or asymptotic covariance
matrix is available. In practical situations, however, we only
have a finite number of snapshots of data from which the array
covariance matrix can be estimated. In this paper, we analyze
the finite-data performance of this beamformer with and without spatial smoothing, using first-order perturbation theory. In
particular, we develop expressions for the mean values of the
power gain in any direction of interest, the output power and
the norm of the weight-error vector, as a function of the number of snapshots and the number of smoothing steps. We show
that, in general, the smoothing, in addition to decorrelating the
sources, can also dlleviate the effects of finite-data perturbations.
Next, we reduce the above expressions to the case when no
spatial smoothing is used. These expressions are valid for an
arbitrary array and for arbitrarily correlated signals. For this
case, we also develop an expression for the variance of the
power gain. We simplify these expressions for a single interference case to show explicitly how the SNR, spacing of the interference from the desired signal and the correlation between
them influence the beamformer performance.
Simulations are used to verify the usefulness of the theoretical expressions and the results show an excellent agreement
with predicted results.
I. INTRODUCTION
I
N a minimum variance distortionless response (MVDR)
beamformer, the array weights are chosen so as to pass
the desired directional (look direction) signal without any
distortion while maximally rejecting the interfering signals. The only assumption made is that the desired signal
direction is known a priori.
Since the pioneering work of Capon [6], there has been
much activity in the development of optimum arrays for
radar, sonar, communication, etc. Sensitivity of the adaptive arrays to element errors and to those in the look direction has been extensively studied in the recent past.
Zahm [7] was one of the first to note the disastrous effects
of pointing errors on the performance of adaptive arrays.
Manuscript received April 8, 1990; revised August 9 , 1991.
K. J. Raghunath was with the Department of Electrical Communication
Engineering, Indian Institute of Science, Bangalore 50012, India. He is
now with the Department of Electrical Engineering, University of Minnesota, Minneapolis, MN.
V. U. Reddy is with the Department of Electrical Communication Engineering, Indian Institute of Science, Bangalore 560012, India.
IEEE Log Number 9202780.
Cox [8] investigated the effects of mismatch in the look
direction on the performance of two optimum beamformers, one based on the inversion of the noise alone crossspectral matrix and the other based on the inversion of
signal-plus-noise cross-spectral matrix. Vural [9] studied
the effects of system and medium perturbations (from assumed conditions) on the performance of adaptive arrays.
Godara [ 101 presented an analysis of the effects of random
errors in the weight vector and steering vector on the performance of two beamformers, the same as those considered by Cox.
This problem, i.e., sensitivity of the beamformer to errors of various kinds, was addressed by Hudson [ l l ] ,
Compton [12], and several others (see Godara [lo] for
more references), and more recently, by Jablon [13]. All
these authors, however, do not address how the performance of the beamformer depends on the data size, i.e.,
the number of snapshots. The only papers (to our knowledge) which address this specific problem are those of
Reed et al. [14] and Boroson [15]. Their analysis, however, assumes that the weight vector computed from one
set of data operates on an independent set of data. Also,
the main thrust of their analysis is directed to the case
where the weight vector is estimated from the noise-alone
matrix inverse, and the interferences are uncorrelated with
the look-direction signal.
A factor which severely affects the performance of the
MVDR beamformer even in the asymptotic case, is the
correlation between the look-direction signal and the interferences. Not only does the beamformer fail to form
deep nulls in the direction of the coherent interferences,
but also the desired signal can be partially or fully cancelled in the output of the beamformer. This signal cancellation phenomenon in the presence of coherent interferences was recently analyzed by Reddy et al. [4] and
Paulraj et al. [17]. For the case of a uniform array, however, spatial smoothing has been proposed to alleviate the
problems due to correlation [ 181. In [4], Reddy et al. have
shown that the effective correlation between the desireddirectional signal and the interfering signals decreases
asymptotically with progressive smoothing thereby reducing the effects of correlation.
The analysis in [4] and [17] has been camed out for
asymptotic case. In this paper, we analyze the performance of the MVDR beamformer in finite data with and
1053-587X/92$3.00 @ 1992 IEEE
I
I
IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 40, NO. 11, NOVEMBER 1992
2136
where tr (.) denotes the trace operator. Now taking the
summation over i inside the trace operator and using (B.2),
we finally obtain
1
E[641
=
c c 411 TI
K
K
u = l 1,=1
(B .6)
where
TI
=
tr (R-’RU,,).
Similarly, we can find the expectation of the other terms
in (3.12) and (3.19).
1121 R. T. Compton, Jr., “Pointing accuracy and dynamic range in a
steered beam array,’’ Trans. Aerosp. Electron. Syst., vol. AES-16,
pp. 280-287, May 1980.
[I31 N . K. Jablon, “Adaptive beamforming with imperfect arrays,” Ph.D.
dissertation, Dep. Elec. Eng., Stanford Univ., Aug. 1985.
[I41 L. S. Reed, J. D. Mallet, and L. E. Brennan, “Rapid convergence
rate in adaptive arrays,” IEEE Trans. Aerosp. Electron. Syst., vol.
AES-10, pp. 853-863, NOV. 1974.
[I51 D. M. Boroson, “Sample size considerations for adaptive arrays,”
IEEE Trans. Aerosp. Electron. Syst., vol. AES-16, pp. 446-451, July
1980.
[I61 D. R. Farrier, D. J. Jeffries, and R. Mardani, “Theoretical performance prediction of the MUSIC algorithm,” Proc. Ins?. Elec. Eng.,
vol. 135, pt. F, no. 3 , pp. 216-224, June 1988.
[I71 A. Paulraj, V. U. Reddy, and T. Kailath, “Analysis of signal cancellation due to multipath in optimum beamformers for moving arrays,” IEEEJ. Ocean. Eng., vol. OE-12, pp. 163-172, Jan. 1987.
[ 181 T. J. Shan and T. Kailath, “Adaptive beamforming for coherent signals and interferences,” IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-33, pp. 527-536, June 1985.
[I91 A. Paulraj, V. U. Reddy, T . J. Shan, and T . Kailath, “Performance
analysis of the MUSIC algorithm with spatial smoothing in the presence of coherent sources,” in Proc. IEEE MILCOM Con$ (Monterey, CA), Oct. 1986.
ACKNOWLEDGMENT
The authors would like to thank the referees for their
very careful reading of the manuscript and their many
useful suggestions, which greatly increased the quality and
readability of this paper.
Kalavai J. Raghunath (S’87) received the B.E.
degree in electronics and communication engineering from Osmania University, Hyderabad, India, in 1988, and the M.S. (Engg.) degree in electrical communication engineering from the Indian
Institute of Science, Bangalore, India, in 1990.
Currently, he is working towards the Ph.D. degree in electrical engineering at the University of
Minnesota, Minneapolis. His research interests are
in the areas of VLSI algorithms and architectures
for signal processing, neural networks, and statis-
REFERENCES
[ I ] M. Kaveh and A. J. Barabell, “The statistical performance of the
MUSIC and minimum-norm algorithms,” IEEE Trans. Acoust.,
Speech, Signal Processing, vol. 34, pp. 331-341, Apr. 1986.
[2] D. R. Brillinger, Time Series: Data Analysis and Theory. New York:
Holt, Rinehart and Winston, 1975.
131 J. H. Wilkinson, The Algebraic Eigenvulue Problem. New York:
Oxford University Press, 1965.
141 V. U. Reddy, A. Paulraj, and T . Kailath, “Performance analysis of
the optimum beamformer in the presence of correlated sources and its
behavior under spatial smoothing,” IEEE Trans. Acoust., Speech,
Signal Processing, vol. ASSP-35, pp. 927-936, July 1987.
151 K. J. Raghunath, “Performance analysis of MVDR beamformer in
finite data: With and without spatial smoothing,” M.Sc.(Eng) thesis,
Dep. Elec. Commun. Eng., Indian Institute Sci., Mar. 1990.
161 J. Capon, “High-resolution frequency-wavenumber spectrum analysis,” Proc. IEEE, vol. 57. pp. 1408-1418, Aug. 1969.
[7] C. L. Zahm, “Effects of the errors in the direction of incidence on
the performance of an adaptive array,” Proc. IEEE, vol. 60, no. 8,
pp. 1008-1009, Aug. 1972.
181 H . Cox, “Resolving power and sensitivity to mismatch of optimum
array processors,” J. Acoust. SOC. Amer., pp. 771-785, vol. 50, Sept.
1973.
[91 A. M. Vural, “Effects of perturbations on the performance of optimumladaptive arrays,” IEEE Trans. Aerosp. Electron. Syst., vol,
AES-15, pp. 76-87, Jan. 1979.
[lo] L. C. Godara, “Error analysis of the optimal antenna array processors,” IEEE Trans. Aerosp. Electron. Syst., vol. AES-22, pp. 395409, July 1986.
[ I l l J. E. Hudson, Adaptive Array Principles. New York: Peter Peregrinus and IEEE, 1981,
tical signal processing.
V. Umapathi Reddy (S’66-M’70-SM’82) received the B.E. and M.Tech degrees in electronics and communication engineering, from Osmania University and the Indian Institute of
Technology (IIT), Kharagpur, in 1962 and 1963,
respectively, and Ph.D. degree in electrical engineering from the University of Missouri in 197 1.
He was an Assistant Professor at IIT, Madras,
during 1972-1976 and Professor at IIT, Kharagpur, during 1976-1979. During 1979-1982, he
was a Visiting Professor at the Department of
Electrical Engineering, Stanford University. In April 1982, he joined Osmania University as a Professor, and was the Founder-Director of the Research and Training Unit for Navigational Electronics, funded by the Department of Electronics, Government of India. Since April 1988, he has
been with the Indian Institute of Science, Bangalore, as a Professor of Electrical Communication Engineering. He has served as a consultant in signal
processing to Avionics Design Bureau of Hindustan Aeronautics Limited,
Hyderabad, and to Central Research Laboratory, Bharat Electronics Ltd.,
Bangalore. His recent research interests are in sensitivity analysis, adaptive
algorithms, adaptive arrays, and neural networks.
Dr. Reddy is a Fellow of the Indian Academy of Sciences and Indian
National Academy of Engineering, and Fellow of the Institution of Electronics and Telecommunications Engineers (IETE), India. He received the
S. K . Mitra Memorial Award (1989) from IETE for the best research paper.
Download